From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugzilla@gcc.gnu.org>
Received: by sourceware.org (Postfix, from userid 48)
	id 819463858D33; Thu,  4 May 2023 22:59:23 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 819463858D33
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1683241163;
	bh=7urDS3jUOraz2g4iEOlMVdmGlYf8a6hE0uv5k6U586w=;
	h=From:To:Subject:Date:In-Reply-To:References:From;
	b=A57huJkXUFiPm/NK9sqndtjeKvcD4Zp2AvhaSP5Y3bbGg/ZIFUb1TWdrdrciikEa5
	 zIYD2BcM644KHWvAluCrH+jwLd9a4YopvcKPIj8Grb/NE5B93u4UnSNQ19NZjRMNgL
	 s2wU9ScAfrQFaW81QxrV1223cEijtMj+fZuRwaZU=
From: "pinskia at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/109732] [14 regression] gcc miscompiles
 iterator comparison on nlohmann_json since r14-204-gf1f5cbaa3f716f
Date: Thu, 04 May 2023 22:59:22 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: tree-optimization
X-Bugzilla-Version: 14.0
X-Bugzilla-Keywords: patch, wrong-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: pinskia at gcc dot gnu.org
X-Bugzilla-Status: ASSIGNED
X-Bugzilla-Resolution: 
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: pinskia at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 14.0
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-109732-4-EszkStWDHA@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-109732-4@http.gcc.gnu.org/bugzilla/>
References: <bug-109732-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
List-Id: <gcc-bugs.sourceware.org>

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D109732

--- Comment #14 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Figured out a C testcase:
```
[[gnu::noipa]]
_Bool f3(_Bool a)
{
        if (a=3D=3D0)
          return 0;
        else
          return a;
}
```
Still need: `-fdisable-tree-fre1` though because fre is able to figure out =
that
the return value is always a. We don't need to disable ethread any more sin=
ce
cfgcleanup will not remove the forwarding BB as it has a predict statement =
in
it.

There are actually two conditions that need to happen to get the wrong code.
First is the order of the phi and then also the order of the edges. The abo=
ve
is enough to get both.=